Rapid Assessment Can Detect Delirium in Hospital Patients in Seconds
October 11th, 2016
UNIVERSITY PARK, PA – In less than a minute, hospital clinicians can determine with a high rate of accuracy whether an older patient is suffering from delirium.
So says a study published recently in the Journal of Hospital Medicine. The assessment instrument consists of just two questions, according to the study team.
"Delirium can be very costly and deadly – and with high-risk patients, time matters," said lead author Donna M. Fick, PhD, of Penn State. "Our ultra-brief two-item bedside test for delirium takes an average of 36 seconds to perform and has a sensitivity of 93%."
Co-author Edward R. Marcantonio, MD, of the Harvard Medical School, recently developed the 3D-CAM, a 3-minute confusion assessment method, to help quickly identify patients with delirium. In developing the new, more concise instrument, Fick said she and Marcantonio wanted to develop an assessment that would be easier to use at the bedside and take less time for a busy nurse or hospitalist.
"We started by looking for one question that could detect delirium, but we could only get 83% sensitivity, which is not good enough," Fick said.
Eventually, the study team settled on two questions that proved to have 93% sensitivity in identifying delirium: Asking what day of the week it was and to recite the months of the year backwards.
If a patient failed to answer those two questions correctly – indicating a strong possibility of delirium – the 3D-CAM was administered.
The 201 study participants, 62% female, were patients at an academic medical center with a mean age of 84. Of those, 21% patients were found to have delirium based on the assessment. The two-item test identified 48 as possibly delirious – 42 were identified correctly, with six false positives.
Researchers note that the item with the best test characteristics was the “months of the year backwards” question with a sensitivity of 83% and specificity of 69%.
Before the test can be widely recommended, however, "These results still need to be validated, with a very large sample,” Fick said.